Anna Lekova

Introduction: Despite strong evidence that repetitive home-based rehabilitation improves functional recovery after stroke, current delivery models still show gaps in continuity of care and patient engagement. AI-driven Embodied Conversational Agents (ECAs) could provide personalized home support through natural-language guidance on prescribed exercises, reinforcement of neuroplasticity, clarification of therapeutic principles, and motivational support. However, clinical deployment remains challenging. Many robotic platforms lack real-time interaction capabilities such as speech processing, gesture execution, and attention tracking, while Large Language Models (LLMs) may produce factual errors or inconsistent responses. Early development is also constrained by limited access to real users due to practical and ethical considerations. Methods: To address these challenges, we propose a Design-Based Research methodology for human–AI co-design and evaluation of ECAs (co-AI DBR), where generative AI facilitates iterative cycles of design, testing, and refinement. Co-AI DBR combines synthetic patient generation with real-code execution to simulate, emulate, and evaluate the ECA platform and its LLM-based conversational pipeline. To validate the method in a post-stroke rehabilitation context, a virtual ECA was first tested with synthetic patients to assess technical implementation and accuracy of LLM responses. A pilot deployment using the Furhat robot as an ECA was then conducted with patient relatives and rehabilitation professionals to evaluate the voice interface and augmented communication. Results: LLM responses to questions from real participants showed higher lexical diversity (MTLD ≈ 134 vs. 93.9) and lower repetition (Yule’s K ≈ 66.8 vs. 115.4) than responses to synthetically generated questions. Responses remained factually consistent, with no contradictions and complete gender invariance, although slightly lower hapax rates were observed (88.8% vs. 99.4%). Usability scores were higher among relatives (M = 86.67) than professionals (M = 72.50), while Intrinsic Motivation Inventory scores indicated similarly high motivation in both groups (M = 6.32 vs. 6.12). Discussion: The results suggest that co-AI DBR can support early design and evaluation of ECAs when direct patient testing is limited. By combining synthetic patient generation with real-code execution, generative AI supports iterative knowledge building during the prototyping and refinement of LLM-based ECAs. This methodology enables the practical development of ECA to support home-based post-stroke rehabilitation.

Advancements in assistive technologies (ATs), particularly those powered by Artificial Intelligence (AI), have significantly enhanced the quality of life and independence of individuals with various disorders, thus fostering progress in education, employment, social engagement and well-being. Despite these advancements, significant challenges remain in the implementation of such technologies in various domains, emphasizing the necessity for deeper exploration of their future development. This study examines the future trends and perspectives of ATs, providing valuable insights from a national context and comparing the results with those of international Delphi research. Using a two-round Delphi method, we engaged a panel of 23 experts from Bulgaria, representing disability organizations, academia and practice. Ten future-oriented projections were assessed, while participant demographics and attitudes toward ATs were also analyzed. The survey’s statistical findings indicate a moderate consensus among the Bulgarian experts. They demonstrate cautious optimism about the use of ATs, especially AI-driven technology, for individuals with disabilities. The results are similar with those of the international study, with only few differences, highlighting the shared perspectives of experts at both national and international levels. The research offers valuable insights into emerging trends in AI-driven ATs and provides valuable knowledge for policymakers, researchers and developers seeking to align innovation with the expectations of diverse stakeholders.

Currently, high-tech assistive technologies (ATs), particularly Socially Assistive Robots (SARs), virtual reality (VR) and conversational AI (ConvAI), are considered very useful in supporting professionals in Speech and Language Therapy (SLT) for children with communication disorders. However, despite a positive public perception, therapists face difficulties when integrating these technologies into practice due to technical challenges and a lack of user-friendly interfaces. To address this gap, a design-based research approach has been employed to streamline the integration of SARs, VR and ConvAI in SLT, and a new software platform called “ATLog” has been developed for designing interactive and playful learning scenarios with ATs. ATLog’s main features include visual-based programming with graphical interface, enabling therapists to intuitively create personalized interactive scenarios without advanced programming skills. The platform follows a subprocess-oriented design, breaking down SAR skills and VR scenarios into microskills represented by pre-programmed graphical blocks, tailored to specific treatment domains, therapy goals, and language skill levels. The ATLog platform was evaluated by 27 SLT experts using the Technology Acceptance Model (TAM) and System Usability Scale (SUS) questionnaires, extended with additional questions specifically focused on ATLog structure and functionalities. According to the SUS results, most of the experts (74%) evaluated ATLog with grades over 70, indicating high acceptance of its usability. Over half (52%) of the experts rated the additional questions focused on ATLog’s structure and functionalities in the A range (90–100), while 26% rated them in the B range (80–89), showing strong acceptance of the platform for creating and running personalized interactive scenarios with ATs. According to the TAM results, experts gave high grades for both perceived usefulness (44% in the A range) and perceived ease of use (63% in the A range).

Advancements in portable EEG headsets have accelerated Brain-Computer Interface (BCI) applications, particularly in smart home automation and user well-being. However, few BCI platforms support cross-disciplinary users with limited programming skills. To address this, we propose a visual, node-based BCI programming framework using BrainFlow and the OpenBCI toolkit within Node-RED. This approach enables users to stream, process, and extract EEG features, while leveraging BrainFlow APIs as backend child processes to ensure adaptability and seamless updates. The framework was validated through a case study with 14 participants, where an OpenBCI headset was used to control a robotic arm. Results revealed that specific bursts in Event-Related Synchronization (ERS) within certain frequency bands were crucial for attention-based control. θ (theta) and γ (gamma) frequency bands in the frontal lobe were highly significant, as these regions are associated focus and decision-making. Similarly, high β (beta) activity in the left central and parietal lobes demonstrated a strong correlation with motor control and sustained attention. The study also evaluated the BrainFlow 'Mindfulness' metric to assess mental state during task engagement. The average value of this metric was 0.31, with a standard deviation of 0.11, indicating a moderate relative variability with a coefficient of variation ≈ 0.364. The results also highlight the key electrodes and frequency bands involved during attention and concentration, emphasizing the potential of using EEG-based metrics and ERS burst patterns as reliable neural markers to distinguish these states.

Background: Among all professions, teaching is significantly affected by psycho-social risks with approximately 33.33% of educators reporting work-related fatigue. Suggestopedia, an effective pedagogical approach developed in Bulgaria, claims to induce positive psychological and cognitive benefits in both teachers and students. In order to gather scientific evidence, given the above statement, we designed a methodology to detect fatigue in Suggestopedia teachers based on neurocognitive analysis and psychological assessment. Methods: An increase in the EEG theta and alpha band powers is considered among the most reliable markers of fatigue. The proposed methodology introduces a robust framework for fatigue analysis. Initially, the changes in EEG band powers using the resting state EEG activity before and after teaching are measured. Subsequently, validated psychological questionnaires are used to gain subjective feedback on fatigue. The study participants include a control group (traditional teachers) and the test group (suggestopedia teachers) to assess whether suggestopedia practice mitigates fatigue among teachers. Observations: In a pilot study, the EEG data was analyzed by evaluating the interrelations between EEG bands and the alpha–beta ratio. The results of the proposed study are expected to provide comprehensive analysis for the fatigue levels of teachers. In future research, our goal is to position the described methodology as a robust approach for evaluating cognitive and emotional states.

The humanoid robot NAO is widely used in therapy scenarios for children with neurodevelopmental disorders, however it has poor speech recognition and dialog based on a predefined lexicon that results in limited vocabulary and limited number of predefined dialog scenarios. The integration of Conversational Artificial Intelligence (AI) in NAO can significantly enhance and expand the robot's capabilities for intensive speech and listening exercises for children with speech and language disorders. Applying design-based research, the authors present the ongoing effective iteration in the building-testing cycles of a software architecture for Conversational AI in the robot NAO by integrating different AI cloud services for NLP into NAO's native software. Examining the aims and methods of stakeholders interested in integrating Conversational AI in NAO for speech and language therapy, we revealed several technical and ethical challenges. These challenges were successfully addressed with solutions implemented in the Node-RED platform, such as achieving more accurate and speech recognition, generating human-like text based on a context, and implementing multilingual text-to-speech synthesis. Additionally, implementation raises ethical considerations for both developers and therapists, especially regarding the assessment of risks linked with AI systems. We followed the guidelines set in the European AI Act, which categorizes AI systems according to their associated risk levels.

In this paper we have identified several good/promising practices from the literature of using Social Humanoid Robots (SHR) as which can contribute to the education, social integration and improvement of lifestyle of individuals with Autism Spectrum Disorder (ASD). Our interest is on the commercial SHR currently available on the market. We provide analysis and evaluation of the effectiveness, advantages and drawbacks, accessibility, user feedback and expert opinions, ethical considerations and future directions. The analysis proof that SHR are promising computer based Assistive Technology (AT) for individuals with ASD.